Using Agent Studio Agent Studio provides a low-code to high-code environment to build, test, and deploy multi-agent workflows for generative AI applications. Agent Studio OverviewCloudera AI Agent Studio is a versatile low-code to high-code platform for building, testing, and deploying multi-agent workflows. Key features of Agent StudioCloudera AI Agent Studio capabilities include low-code workflow orchestration, custom tool extension, model integration, and Phoenix observability for production-ready AI agents.Shared responsibility model for security in Cloudera AI in Agent StudioCloudera AI Agent Studio runs within the Cloudera AI Workbench. Security responsibilities are shared between Cloudera as the platform and product vendor and you as the customer deploying and operating Agent Studio.Use cases of Agent StudioAgent Studio provides the orchestration environment for building goal-oriented AI systems that automate multi-step workflows, including DevOps, data processing, and compliance screening.Launching Agent Studio within a ProjectAgent Studio is compatible with Cloudera AI Inference service and enabling you to build, test, and deploy multi-agent workflows.Deployment of Agent StudioDeploy Agent Studio in the Cloudera AI Workbench to manage machine learning agent deployments.Migrating WorkflowsThis section details the steps to transfer a workflow template between Cloudera AI Agent Studio instances, covering both export from the source and import into the target instance.Supported inference service providersEnterprise inference service providers connect with Agent Studio to facilitate Large Language Model (LLM) operations and API calls.Agentic workflowsAgent Studio enables the creation of complex agentic workflows, which integrate multiple agents and tools, often using manager agents to orchestrate the entire process. Guardrails in Agent StudioGuardrails in Agent Studio protect your confidential information by blocking unauthorized data transmission to Large Language Models (LLMs) during workflow runs.Task planning: Streamlining agent workflowsTask planning is a workflow feature that enables an agent to manage complex, multi-step projects by generating its own structured to-do list during a run.Native function calling overviewAgent Studio runs all agents and workflows on native function calling, also known as tool calling. This replaces the traditional text-based parsing methods to achieve deterministic, reliable execution of enterprise AI agents.Auditing in Agent StudioAgent Studio provides comprehensive, granular built-in auditing operating at two separate infrastructure levels, on resource management level, and on workflow level. This governance architecture ensures absolute accountability across both organizational configuration modifications and operational executions without adding management overhead.Secure Tool Execution and DevelopmentSecure Tool Execution and Development is a framework designed to mitigate security risks such as credential exposure and data leakage by running user-defined tools in isolated environments instead of a shared runtime.Installing Agent StudioThe following section outlines the steps required to configure and deploy the Agent Studio application within Cloudera AI, including runtime setup and environment configuration.Service Account usage in Agent StudioAgent Studio uses a dedicated Service Account to deploy production or shared workflows instead of a personal user account.Register models in Agent StudioManage Large Language Models (LLMs) within Agent Studio, by registering new models, the supported model providers, and by validating already registered models for proper functioningUser access management with Role-Based Access Control (RBAC)Role-Based Access Control (RBAC) in Agent Studio secures access to workflows, models, and tools by mapping Cloudera AI Workbench project permissions to user roles.Managing workflow EvaluationsThe Evaluations feature offers a comprehensive set of diagnostic and quality-assurance tools designed to measure the performance, accuracy, and safety of your Agentic Workflows. This feature serves the purposes of assessing workflows during the development phase within AI Studio and enabling you to audit historical workflow runs in deployed workflows.Deploying workflows as Model EndpointsThe Cloudera AI Agent Studio deployment system transforms AI workflows into production-ready endpoints that operate as an independent services.Model Context Protocol (MCP) integrationMCP enhances AI workflows by enabling seamless integration with external systems and tools. It provides a standardized approach for secure communication, allowing AI agents to interact efficiently with diverse services while maintaining flexibility and control within Cloudera AI Agent Studio.Monitoring in Agent StudioThe Monitoring feature in Agent Studio integrates the Phoenix observability tool to provide deep insights into workflow execution and trace data for asynchronous test and production workflows.Using inbuilt toolsUse inbuilt tools in Agent Studio as the foundational building blocks for AI agents and workflows to interact with external systems, such as databases and APIs, and perform tasks that are beyond standard LLM capabilities.
